drd_balance(8)

Index for
Section 8
Alphabetical
listing for D
Bottom of
page
drd_balance(8)
NAME
  drd_balance - Distributed raw disk (DRD) load balancing utility (provided
  on Production Server configurations only)

SYNOPSIS
  /usr/sbin/drd_balance [flags]

FLAGS
  -d	    Displays additional debugging information.

  -D	    Displays full debugging information.

  -h	    Displays help message.

  -i num    Sets the polling interval to the specified number of seconds.  To
	    avoid excessive service relocations, do not set the polling
	    interval to less than a few minutes.  In addition to degrading
	    system performance, short polling intervals do not provide enough
	    time for the drd_balance utility to determine accurate usage
	    patterns.

  -p num    Specifies the threshold at which the drd_balance utility begins
	    to suggest the relocation of the DRD services.  The num parameter
	    represents the percentage of all DRD I/O operations performed by
	    the local node that are remote DRD I/O operations.	The value of
	    num should be in the range of 50 to 100.

	    Use the -p flag when most remote DRD I/O operations are started
	    within the available server environment (ASE) to which this
	    member system belongs, or are started about equally between this
	    ASE and a remote ASE.  When the specified threshold is exceeded,
	    the drd_balance utility considers all remote I/O requests
	    initiated within the cluster when making relocation suggestions.

  -P num    Specifies the threshold at which the drd_balance utility begins
	    to suggest the relocation of DRD services.	The num parameter
	    represents the percentage of all DRD I/O operations performed by
	    the local node that are remote DRD I/O operations.	The value of
	    num should be in the range of 1 to 100.

	    Use the -P flag when a node outside of the ASE to which this
	    member system belongs is starting most remote DRD I/O operations.
	    When the specified threshold is exceeded, the drd_balance utility
	    considers only remote I/O requests initiated within the current
	    ASE when making relocation suggestions.

  -r	    Causes the drd_balance utility to attempt the service relocations
	    it determines when the -p or -P flag is specified.	If you do not
	    specify the -r flag, the drd_balance utility displays the
	    suggested relocations and does not attempt to perform them.

  -s num    Adds the specified DRD service to the list of services the
	    drd_balance utility is monitoring for potential relocations.  The
	    num parameter represents the numeric portion of the DRD service's
	    name.  To explicitly monitor more than one DRD service,  use this
	    option multiple times on the command line.

	    By default, the drd_balance utility monitors all DRD services.

	    You cannot use the -s and -S flags on the same command line.

  -S num    Adds the specified DRD service to the list of services the
	    drd_balance utility is not monitoring for potential relocations.
	    The num parameter represents the numeric portion of the DRD
	    service's name.  You cannot use the -s and -S flags on the same
	    command line.

  -t	    Disables the printing of a timestamp when the drd_balance utility
	    performs each relocation.

DESCRIPTION
  The drd_balance utility analyzes DRD I/O usage patterns in a Production
  Server configuration. It does so by gathering statistics that describe the
  number of DRD read and write requests that are started by the local node
  and by remote nodes.	 If the node running the utility is servicing more
  remote requests than local requests for a particular DRD service, the
  utility determines if another cluster member can more efficiently provide
  the service.

  The drd_balance utility polls the DRD subsystem at a tunable time interval
  for operation counts and a list of cluster members issuing remote DRD
  service requests.  Because it analyzes DRD service usage statistics only
  for DRD services provided by the node it is running on, it can only
  relocate services from the current node to another node within the same
  ASE.	To optimally balance DRD service placement, have a separate instance
  of the drd_balance utility active on each cluster member.

  The drd_balance utility can relocate a DRD service only to other nodes
  within the same ASE.	It displays messages indicating when a majority of
  remote requests for a local DRD service are being initiated by a cluster
  member in another ASE.  Although the utility cannot perform the relocation
  in this situation, the messages are an indication that DRD performance
  could benefit from a disk configuration change.

				     Note
       Because DRD I/O usage patterns vary over time, the suggested service
       relocation is not guaranteed to improve performance.

  If the DRD service consists of more than one DRD device or a DRD device
  that is a Logical Storage Manager (LSM) volume, the drd_balance utility
  does not attempt to relocate the service.

EXAMPLES
  The following command enables relocations of all DRD services:

       % drd_balance -r

  The following command enables relocations and sets the polling interval to
  five minutes (300 seconds):

       % drd_balance -r -i 300

  The following command enables relocations for the DRD services drd1, drd2,
  drd8, and drd11:

       % drd_balance -r -s 1 -s 2 -s 8 -s 11

  The following command enables relocations for all DRD services except drd1
  and drd11:

       % drd_balance -r -S 1 -S 11

RELATED INFORMATION
  drd(7), asemgr(8)
Index for
Section 8
Alphabetical
listing for D
Top of
page